Brown bullhead catfish melanoma represents a novel transmissible cancer

In 2012, anglers and biologists began observing a high proportion of brown bullhead (A. nebulosus, a type of catfish) with raised black lesions in Lake Memphremagog, a freshwater lake in Vermont, USA, and Quebec, Canada¹ (Fig. 1). A study determined these raised black lesions to be malignant melanomas, on the basis of histopathological analysis, and observed these lesions in 23–37% of brown bullhead sampled between 2014 and 2017 (ref. ¹). Melanomas have been reported from brown bullheads in other bodies of water, but generally at considerably lower incidence¹. Brown bullheads are widely distributed across Eastern North America and are considered an indicator species of water quality¹⁴. The high incidence of cancer in Lake Memphremagog could indicate a local cancer-causing contagion or contaminant. One hypothesis is that extensive flooding in 2011 from tropical storm Irene led to reduced water quality of the lake due to runoff pollution. Monitoring changes to water quality is of great concern to biologists and the public, as this lake is the source of drinking water for more than 175,000 people¹⁵.

To identify the cause of the melanomas, researchers initiated multiple studies, including genomic analyses to determine whether there were differences between diseased and healthy fish. Preliminary results indicated that the genomics of the tumours were considerably different from those of the healthy tissue. This led to the hypothesis that this cancer may be a transmissible cancer, because this genomic observation is consistent with known clonal cancer lineages that spread through a population by engrafting on new organisms like a contagious parasite².

A transmissible cancer originates as a conventional cancer in a single founder individual, with cancer cells then metastasizing to infect new hosts—the cancer cells themselves are the agent of transmission¹⁶. Thus far, naturally occurring transmissible cancers have been observed in only three types of animals: dogs, Tasmanian devils and several species of bivalve². In dogs, canine transmissible venereal tumour is a single lineage that arose thousands of years ago and is usually survived by the host^4,17. By contrast, Tasmanian devil facial tumour disease is only decades old, is generally lethal for the host and has arisen independently twice^5,18,19. Bivalve transmissible neoplasia is a leukaemia-like cancer, arising and spreading independently at least ten times across at least ten bivalve species, in some cases spreading within a different host species than the one in which it originated^{6,7,8,9,10,11,12,13}. Although no transmissible cancers have been documented in any fish species, a researcher in the early 1900s observed an outbreak of black tumours in brown bullheads from a Massachusetts pond²⁰. They subsequently induced small tumours in naive fish through contact with tumour cells, suggesting that the cancer was caused by a contagion, but could not identify a transmissible agent at the time²⁰. Microbiome analyses conducted thus far have not revealed any viruses or other microorganisms unique to the diseased bullhead in Lake Memphremagog (Supplementary Information 7 and Supplementary Tables 10 and 11), leaving the possibility that the transmitting agent may be the cancer cells themself.

An indicator of a transmissible cancer is when cancer genomes are more closely related to each other than they are to the genome of the host they are found in—across infected individuals, the transmissible cancers will appear clonal^4,5,6,21 (Fig. 1f). Moreover, large numbers of genetic variants are shared among the cancer cells from genetically distinct hosts due to their common ancestor—a single founder animal in which the cancer originally initiated. Alternatively, conventional cancers will closely match their host genomes, and most mutations will be unique to that individual tumour, even if the cancer was caused by a contagious agent such as a virus, with few if any mutations shared across independent cancers.

Here we test the hypothesis that malignant melanoma in Lake Memphremagog brown bullheads represents a clonal cancer lineage. We genomically characterized normal tissue from unaffected fish and paired normal and tumour tissues from fish with histologically confirmed melanistic skin lesions. We chose fish with a range of lesion severity that includes non-raised lesions (Fig. 1b,c) with no evidence of tumour formation but proliferations of melanocytes in the dermis and epidermis, to raised black lesions (Fig. 1d,e) of developed melanomas where the normal skin structure was no longer apparent, with melanocytes invading the underlying muscle and, in some cases, metastasizing to other organs. We found extensive genomic evidence to suggest these melanistic lesions represent a transmissible cancer, to our knowledge, the first known occurrence in a fish species and the first found in a freshwater aquatic setting.

We tested the hypothesis that black melanistic lesions found on A. nebulosus individuals in Lake Memphremagog are derived from a clonal lineage of tumour cells independent from the host fish. We built and used a mitochondrial draft genome (NCBI: PV874414.1) and a nuclear draft genome (NCBI: GCA_051624135.1), assembled and annotated from long- and short-read sequences from fish HL4 (Supplementary Information 2 and 3), as the references for all mitochondrial and nuclear genome analyses. After whole-genome sequencing, we analysed mitochondrial and nuclear DNA sequences of normal brain tissue from ten unaffected reference fish populations from Vermont (n = 7), New Hampshire (n = 2) and Maine (n = 1), and normal brain tissue from 84 unaffected fish in Lake Memphremagog, and 19 paired tumour skin and normal skin or brain tissues (n = 9, collected in 2019, and n = 10, collected in 2023, respectively) from Lake Memphremagog (Fig. 1a and Supplementary Tables 1 and 2). We also took advantage of a 2015 RNA-sequencing dataset from normal skin tissue from an additional reference population of unaffected fish from Ticklenaked Pond in Vermont (n = 4), and normal and melanistic skin tissues from unaffected (n = 5) and affected (n = 9 tumour–normal pairs) fish from Lake Memphremagog (Supplementary Table 3). For the initial analyses, we focused on mitochondrial DNA due to its higher read depth and the simplicity of analysing haploid DNA in mixed samples but subsequently analysed nuclear-DNA single-nucleotide variants (SNVs) and structural variants (SVs) as well.

We identified heteroplasmy—that is, multiple distinct mitochondrial haplotypes within a tissue—in a subset of normal skin and brain samples, and in all tumour samples (Fig. 2 and Extended Data Fig. 1). Although heteroplasmy was present in a large portion of normal tissues (11% skin and 12.6% brain samples), it was usually a single SNV and never more than four SNVs in a single sample. We interpret these SNVs in normal tissues as somatic mosaicism; genetically distinct cell populations within an individual arising either from post-zygotic mitochondrial mutations or from maternally inherited heteroplasmy. By contrast, all tumour samples exhibited heteroplasmy and contained nearly 10× more heteroplasmic SNVs (32 to 37 SNV sites relative to normal host tissue), the majority of which were shared by all tumour samples (Fig. 2b and Extended Data Fig. 1). We interpret these to be somatic mutations present in the tumour cells that are not present in non-cancerous tissue captured in the sample, and evidence that tumour mitochondria are divergent from non-cancerous tissue from the same sample.

To test whether divergent tumour mitochondria were due to the relatedness of tumour samples, as expected in the case of a transmissible cancer, we explored the phylogenetic relationships between mitochondrial haplotypes. We generated phased mitochondrial genomes for tissue samples when they displayed heteroplasmy. The resulting maximum-likelihood tree showed a monophyletic clade of tumour mitogenomes (see Fig. 2a for tumour–normal pairs and reference fish; Extended Data Fig. 2 for phased DNA and RNA samples; and Extended Data Fig. 3 for all genotyped DNA and RNA samples) regardless of sampling time (2015, 2019 or 2023) that do not resemble either their healthy host tissue or any other population of fish sampled for this study. This indicates a clonal tumour lineage in Lake Memphremagog that predates 2015.

We next investigated whether the nuclear genome also supported a clonal cancer lineage by calling SNVs in the nuclear genome. As the nuclear genome is diploid and much lower coverage than the mitochondrial genome, we did not have the resolution to phase host and clonal cancer genomes within mixed samples. Instead, we took a conservative approach and analysed SNVs for mixed samples. Using 686,296 nuclear SNV loci across all samples, we built a neighbour-joining phylogenetic tree (see Fig. 3a for tumour–normal pairs and reference fish and Extended Data Fig. 4 for all samples) and observed a monophyletic clade for tumour tissue samples. The tumour tissue clade was more similar to unaffected reference fish from New Hampshire and Maine than to the other fish samples from Lake Memphremagog and reference fish from Vermont. Notably, this relationship was apparent despite tumour samples containing a mixture of host and tumour DNA, which would be expected to make tumour samples appear more closely related to their hosts. This further supports the hypothesis that the tumours represent a clonal lineage and also indicates that the lineage originated outside of Lake Memphremagog.

We next identified variants found only in tumour samples or only in paired host brain tissue. We identified 245,189 tumour-specific SNV sites, and 61,699 brain-specific SNV sites across the 16 tumour–normal pairs (Fig. 3b). In addition to finding 3.9× more tumour-only sites relative to normal-only sites, the distributions of SNV sites shared across fish tissues is very different. The distribution of these SNVs across tumour samples shows that the majority (59%) is shared by at least 14 fish (18.8% in 14 fish, 20.6% in 15 fish and 19.6% in 16 fish). By contrast, the majority (83.4%) of brain-only SNV sites is found in only a single fish with none shared by more than 14 normal samples. We find a similar pattern when calling tissue-specific variants with a specialized somatic mutation caller, Mutect2, on the 6 paired tumour–brain samples with the greatest sequencing depth, finding 78.9% of brain-only variants (compared with 63.3% using the original method) are unique to a single fish, while 69.3% of tumour-only variants (compared with 74.7% using original method) are shared by all six tumours (Extended Data Fig. 5).

To put this in context, conventional cancers that arise independently by somatic mutation can share some tumour-specific variants due to random chance, mutational biases (such as CC > TT sites in melanomas) and selection (in tumour-promoting genes), among other possible mechanisms^22,23. As a reference for how many shared tumour-specific variants to expect between conventional cancers, we analysed SNVs in melanomas in The Cancer Genome Atlas (TCGA)^3,24. Among 463 human melanomas, 94% of SNVs are unique to a single cancer, and only 40 out of 498,648 SNVs (0.008%) are shared by 10 or more cancers (Extended Data Fig. 6a). To match the human data to the 16 bullhead tumours in our SNV set, we calculated the average number of shared mutations across 100 random sampling permutations of 16 TCGA human melanomas. We found that 99.9% of human SNVs were unique to a single cancer in those samplings, and no SNVs were shared by more than 12 melanoma samples in any of the 100 permutations (Extended Data Fig. 6b). This contrasts with bullhead melanoma, in which the majority of tumour-specific SNVs are shared by other tumours (Fig. 3a), highlighting that this level of shared mutation is considerably higher than what would be expected between conventional cancers.

Large-scale variation in the structure of DNA is an important aspect of genomic diversity, particularly in cancer lineages, that can have large effects on cellular function through changes in genomic content and gene expression. We surveyed genomic structural variants (SVs); >15 bp deletions (n = 25,642), duplications (n = 1,313), insertions (n = 7,992) and inversions (n = 792) across tumour and normal samples. Phylogenetic trees generated from these SVs for short- and long-read sequence datasets were generally consistent with the single-nucleotide polymorphism (SNP) tree and all trees resolved the tumour monophyletic clade (Fig. 3b and Extended Data Figs. 7 and 8). The New Hampshire and Maine samples (short reads only) always formed a monophyletic clade that was frequently sister to the tumour clade (for example, short-read insertions and deletions). The Vermont reference sequences tended to form a monophyletic clade (except for short-read insertions and inversions), and clustered with the Lake Memphremagog normal tissues, which tended to form a paraphyletic clade (except for short-read deletions).

For each SV type, we filtered for tissue-specific SVs and counted the number of samples that shared each SV (Fig. 4a–d), as previously described for SNVs. We found vastly more tumour-only SVs shared by all tumour samples than normal-only SVs shared by all paired normal tissues. For example, in short-read data we found 2,572 tumour-specific deletions compared with 166 brain-specific deletions. Of those tissue-specific deletions, 767 (75%) were present in at least 14 tumour samples, but no normal-specific deletions were shared by more than eight normal samples.

Copy-number variation (CNV) profiles (for example, regional duplications and deletions) across the genome showed distinct patterns within and between tumour and normal samples (Fig. 4e–h, Extended Data Fig. 9 and Supplementary Tables 7–9). Tumour tissues shared minimal CNV with paired host tissue (Pearson’s r mean = 0.22 and median = 0.21; Fig. 4f, Extended Data Fig. 9 and Supplementary Table 7), but tumour tissues shared most of their CNV with other tumours (Pearson’s r mean = 0.83 and median = 0.84; Fig. 4i). A similar pattern is seen between normal tissues that share much of their CNV with each other (Pearson’s r mean = 0.73 and median = 0.72; Fig. 4g). There is variation in the relationship between normalized copy-number profiles for different tissue pair comparisons (tumour versus normal in the same fish, tumour versus tumour between fish, and normal versus normal between fish), which is probably due to variation in sequencing depth, host contamination in tumour tissue, and genetic distance between strains and hosts. The general trend of similar CNV profiles between tumours further supports that they originate as a clonal lineage.

Overall, nuclear genomic variant analyses of single-nucleotide and structural variation support a clonal origin for a transmissible neoplasia in A. nebulosus. Moreover, the phylogenetic analyses show that the normal tissue from these tumour afflicted fish is distinct and distantly related to the genomic DNA found in the tumours growing on these fish. The phylogenetic signal is apparent despite the tumour tissue's being a mosaic of clonal tumour and host cells.

Here we show brown bullhead melanoma in Lake Memphremagog is a single clonal cancer lineage, with tumours more closely related to each other than they are to their hosts through several lines of mitochondrial and nuclear evidence, and across multiple timepoints. This makes it to our knowledge the first documented transmissible cancer in fish or in a freshwater ecosystem. To better understand the potential implications of this transmissible tumour on brown bullhead populations, it is necessary to better understand how the cancer originated and the mechanism of spread between individuals. The two major barriers to a cancer spreading between individuals are the ability to repeatedly (1) physically transfer between individuals and (2) evade host immune systems^2,16.

While it is unclear how brown bullhead transmissible melanoma transmits, we can infer possible mechanisms from bullhead biology and the other characterized transmissible cancers. Dogs primarily transmit tumours sexually⁴, Tasmanian devils primarily transmit tumours by biting⁵ and bivalve cancers are believed to transmit through passive seawater transfer^25,26,27. Brown bullhead melanoma has not been documented in fish younger than reproductive age¹, indicating either that there is a long latent period before the cancer is detectable, or that reproduction/ageing itself has a role in infection. There are two potential reasons that reproduction may have a role in tumour transmission: (1) reproductive behaviours; and (2) hormonal changes. Brown bullheads aggregate together in relatively small areas during spawning, with physical interactions that offer the opportunity for cells to transfer directly between individuals²⁸. As observed in the bivalve cancers, the aquatic environment can allow cells to transmit indirectly, in which case they might engraft on a new host fish through the gills, skin or mouth. Furthermore, bullheads are scaleless, benthic fish, making it possible that clonal cancer cells in sediment might transfer to passing fish. Another factor could be weakened host immune systems due to chemical pollutants or hormonal imbalance at the time of spawning. Fish immune responses are modulated by reproductive hormones²⁹, which could make them more susceptible to infection by transmissible tumour cells. Regardless of the mechanism, the rapid spread of melanoma to infect 30% of fish in Lake Memphremagog by 2015, after not being present at noteworthy levels before 2012 (ref. ¹), indicates a high rate of transmission in the population. Future sampling of additional populations of affected brown bullhead will address the age and origin of this clonal cancer, which is also essential for understanding the evolution of this clonal cancer and the way it impacts brown bullhead populations.

Ultimately, the major time-sensitive question to answer is how this transmissible cancer will affect brown bullhead populations. Although the Tasmanian devil tumour is nearly uniformly fatal, contributing to a 90% population decline and threatening extinction³⁰, the dog tumour generally regresses on its own³¹ and bivalve populations persist despite widespread transmissible cancer incidence, although localized mortality events have been documented^32,33. It is currently unclear how the disease has affected brown bullhead abundances in Lake Memphremagog. However, over time, the effect of the cancer on the broader population may be more pronounced. We also do not know the extent to which this cancer lineage may already exist in other populations throughout North America. Observations of brown bullhead with melanomas date back over a century²⁰, which could represent this same lineage, independent transmissible cancer lineages or conventional cancers. Other transmissible cancers range in age from decades (in Tasmanian devils¹⁹) to centuries (in clams³⁴) to millennia (in dogs¹⁷), and have generally spread widely within their host species’ range, even when separated by continents^8,17,35. Broader sampling, and a molecular clock estimate for how long ago the cancer originated, would be informative of the extent this transmissible cancer has spread in brown bullhead. Information about the cancer lethality and transmissibility will be critical to modelling how further spread might affect naive populations and inform the necessity of containment efforts to minimize natural or anthropogenic spread to other water bodies.

This finding that transmissible cancers, previously detected only in dogs, devils and bivalve species, also exist in brown bullheads raises the question of whether other fish species may have transmissible cancers as well. Other fish species have been reported with melanistic lesions, some with putative causes (for example, Xiphophorus hybrids³⁶) and others of unknown cause (such as coral trout³⁷ and smallmouth bass³⁸), which may warrant testing for clonality. Transmissible cancers are a puzzling biological phenomenon that are not expected to exist, but are now confirmed in over a dozen species, with the potential to devastate host populations. As we continue to identify examples of transmissible cancers, it becomes increasingly clear that they may be more common and ecologically important than previously thought. Understanding the origins, transmission and impacts of brown bullhead melanoma will not only inform conservation and management of this species but also broaden our understanding of cancer evolution and host–pathogen dynamics in natural ecosystems.

Field collection

Fish collected for lethal sampling were euthanized with an overdose of buffered MS-222 solution (tricaine methanesulfonate, Syndel). Tissues from A. nebulosus individuals were collected (Fig. 1a) at five timepoints as normal brain or skin tissue (N) or melanistic skin tissue from fish with black lesions (T). A set of samples was collected from normal and tumour skin tissue during June of 2019 from Lake Memphremagog in Hospital Cove (2019, n = 8 (N), n = 8 (T)). A second set of Lake Memphremagog fish were collected in June of 2023 from eight locations in Lake Memphremagog for unaffected fish brain tissue only (N) and affected fish paired brain (N) and tumour tissue (T) genomic analysis (Fig. 1 and Supplementary Table 2); Fitch Bay (FB, n = 6 (N) and n = 8 (paired N and T)), Magog Area (MA, n = 6 (N) and n = 8 (paired N and T)), Hospital Cove (HC, n = 8 (N) and n = 6 (paired N and T)), Sargent’s Bay (SB, n = 6 (N) and n = 6 (paired N and T)), Scott’s Cove (SC, n = 6 (N) and n = 5 (paired N and T)), South Bay/Johns River (JR, n = 6 (N) and n = 6 (paired N and T)) and South Main Lake (SML, n = 6 (N) and n = 6 (paired N and T)). A set of reference fish (Supplementary Table 1) was collected during spring 2022 from Lake Bomoseen (BS, n = 2 (N)), Lake Champlain Larabee’s (LB, n = 2 (N)) and Sandy Point (SP, n = 2 (N)), and the Connecticut river at Hoyts Landing (HL, n = 2 Sados (N)). Additional reference fish (Supplementary Table 1) were collected in 2024 from Maine (ME, n = 1 (N)) and in 2025 from New Hampshire (VP, n = 2 (N)). A set of transcriptome samples collected from 2015 Lake Memphremagog healthy skin and tumour skin, and healthy skin from Ticklenaked Pond were analysed for mitochondrial sequence and the data are presented in Supplementary Information 1 and 5 and Supplementary Table 3.

Normal tissues were sampled, DNA extracted and whole-genome sequenced for all (normal and malignant melanoma) Lake Memphremagog fish included in this study. However, not every fish with a malignant melanoma had a tumour tissue sampled, DNA extracted or sequenced (Supplementary Table 2). Histology was performed on a subset of the fish with melanistic lesions to verify that these tumours are consistent with those in the previous study¹. We note that not all of the melanistic tissues used for genomic analysis were characterized by histology³⁹. If a tumour sample had a low variant allele frequency (>15% tumour mitochondria relative to host mitochondria, see below for details), we resampled and sequenced tumour tissue from the same fish. Collections for this study were performed under the auspices of the Vermont Fish and Wildlife Department.

Tissue processing and nucleic acid sequencing

Tissue samples were collected as soon as possible after euthanasia. All necropsy equipment was butane-flame sterilized between animals. Tissue samples used for DNA analysis at the Vermont Integrated Genomic Resource at the University of Vermont (VIGR; RRID: SCR_021775) were stored without a buffer and placed on ice. The samples were transported to VIGR on wet ice and stored in a −80 °C freezer.

All fish tissue was subsampled under sterile conditions and with equipment disinfected with 10% bleach. The 2022 reference fish were extracted using the DNeasy Blood and Tissue Kit (Qiagen). Tissues were lysed in 180 µl ATL buffer plus 20 µl proteinase K and incubated at 56 °C for 2.5 h with periodic vortexing. The lysates were extracted for DNA using the manufacturer’s protocol and eluted in 200 µl of buffer AE. The Lake Memphremagog, Maine and New Hampshire samples were extracted for DNA using the E-Z 96 Tissue DNA Kit (Omega Bio-Tek) kit but with the following modifications. The samples were lysed in 200 μl TL buffer plus 25 µl proteinase K and incubated at 60 °C overnight with periodic vortexing. The lysates were extracted according to the manufacturer’s protocol and eluted in 150 µl of the elution buffer. All extracted DNA was quantified using the Qubit spectrofluorometer (Thermo Fisher Scientific) using the dsDNA HS kit.

High-throughput sequencing

DNA short-read sequencing

Whole-genome DNA-sequencing libraries were prepared using the NEXTFLEX Rapid XP V2 DNA-Seq kit (Revvity) according to the manufacturer’s protocol with the following modifications. For the reference samples and 2019 Lake Memphremagog samples, 150 ng total DNA was used for input and the libraries were amplified for four cycles of PCR. For the 2023 Lake Memphremagog samples, 10 ng of total DNA was used for input, and the libraries were amplified for 7–9 cycles of PCR using SG UDI Primers (Singular Genomics) rather than Revvity Primer Mix V2. The final libraries were eluted in 5 mM Tris, and the library quality and concentrations were determined using the High Sensitivity DNA Kit on the Agilent Bioanalyzer 2100 and Qubit spectrofluorometer (Thermo Fisher Scientific). The samples in each group were pooled equally on the basis of total mass (ng). The reference fish and the 2019 Lake Memphremagog samples were sequenced as paired-end 150 bp or 250 bp reads on the S2 flow cell on the Illumina NovaSeq 6000 system at the UNH Hubbard Center for Genomics. The 2023 Lake Memphremagog fish were sequenced as paired-end 150 bp reads using the Singular Genomics G4 system at the VIGR and at Singular Genomics.

Long-read sequencing

Libraries for direct whole genome DNA sequencing were prepared for all samples (Supplementary Table 5) using the Native Barcoding Kit (Oxford Nanopore Technologies (ONT)) and were sequenced on the ONT PromethION 2 Solo device using R10.4.1 flow cells according to the manufacturer’s protocol. Long-read sequencing, FAST5 or Pod5 files were base called and filtered (min qscore 7; hac,5mCG_5hmCG), demultiplexed and adapter trimmed using Dorado v.0.7.2 (https://github.com/nanoporetech/dorado).

Details about sequencing and sequence availability are provided in Supplementary Information 2 and Supplementary Tables 4 and 5.

Draft mitochondrial and nuclear genomes

Reference draft mitochondrial and draft nuclear genomes

The draft mitochondrial and nuclear genomes for A. nebulosus were constructed from short-read only (mitochondrial) or short- and long-read (nuclear) sequencing data generated from a healthy male fish (HL4) collected from Hoyt’s Landing, in the Connecticut River (Supplementary Table 1).

The mitochondrial genome was assembled using GetOrganelle (v.1.7.7.1)⁴⁰ from short reads that were quality trimmed and filtered using Fastp (v.0.24.0)⁴¹. The origin was standardized and the mitochondria was annotated using MTGrasp (v.1.1.8)⁴². A full description of results is provided in Supplementary Information 3.

Before the genome build, all HL4 long and short reads were mapped to the HL4 mitochondrial genome using Dorado v.0.7.2 (long reads) or Minimap2 (v.2.28)⁴³ (short reads), and the mitochondrial reads were removed from the raw sequencing data prior to genome assembly using Samtools (v.1.21)⁴⁴.

We built and used a custom nextflow workflow called Pegasus (https://github.com/jaxlub/PEGASUS?tab=readme-ov-file) to build the HL4 draft nuclear genome using a hybrid assembly approach that combines long and short reads (https://github.com/jaxlub/PEGASUS; downloaded 30 September 2024)⁴⁵. The workflow assesses read quality using Nanoplot (v.1.41.0)⁴⁶ and FastQC (v.0.11.9)⁴⁷ and trims reads with Fastp (v.0.23.4)⁴¹. Centrifuge (v.1.0.4)⁴⁸ then removes human, prokaryotic or viral origin reads. Flye (v.2.9)⁴⁹ is used to assemble long-reads into contigs that are polished by Hapo-g (v.1.3.7)⁵⁰ (short reads) and Racon (v.1.5.0)⁵¹ (long reads). The polished contigs are rescaffolded using Ntlink (v.1.3.9)⁵². These scaffolds are repolished again resulting in the de novo genome build. The de novo genome is assessed for quality and coverage using BUSCO (v.5.5.0)⁵³, QUAST (v.5.2.0)⁵⁴ and Mosdepth (v.0.3.6)⁵⁵. The resulting draft genome was annotated with EGAPx v.0.4.1-alpha (https://github.com/ncbi/egapx)⁴³. In addition to the default gene prediction and annotation methods, we used the 2015 A. nebulosus RNA-sequencing data to train the gene prediction model, and the BUSCO actinopterygii_odb10 database, eggNOG_mapper (v.2.1.12)⁵⁶ to improve annotations, and Barrnap (v.0.9)⁵⁷ to predict rRNA (https://github.com/tseemann/barrnap). A description of results is provided in Supplementary Information 4.

Mitogenome variant calling and phasing

Mitochondria were analysed by genotyping samples to generate SNVs, phasing mitochondrial genomes and building phylogenetic trees.

Variant calling and analysis

Short-read data were mapped to the HL4 mitochondrial using Minimap2 (v.2.28)⁴³ then converted, sorted and indexed using Samtools⁴⁴. Haplotyping was done two ways using the GATK (v.4.6.1.0) HaplotypeCaller⁵⁸ using a ploidy of N = 1 or N = 2. Haplotyped samples were genotyped using GATK using Genomics_DBImport and GenotypeGVCFs⁵⁸. Variant calling was done with GATK’s SelectVariants⁵⁸ and hard filtered using the GATK’s recommendations for SNPs (QD < 2.0, QUAL < 30.0, SOR > 3.0, FS > 60.0, MQ < 40.0, MQRankSum < −12.5, ReadPosRankSum < −8.0) or indels (QD < 2.0, QUAL < 30.0, FS > 200.0, ReadPosRankSum < −20.0). Variant calls for the haplotype ploidy analyses of N = 1 and N = 2 were visualized in IGV (v.2.18.5)⁵⁹. Owing to the heterozygosity observed in the N = 2 tumour samples (32 to 37 sites of the 16,513 bp genome; Extended Data Fig. 1), we used the 2 N variant-called dataset for further analyses.

Mitogenome reconstruction

The mitochondrial genomes for all samples with no evidence of heterozygous loci during variant calling were assembled and annotated as described above (n = 99). For the remaining samples, we phased haplotypes on the basis of the minor allele frequency (≥0.10; Extended Data Fig. 2) using custom scripts for R statistical software (v.4.4.2)⁶⁰ (https://github.com/limey-bean/Bullhead_2025/). All tumour samples were additionally phased by subtracting the normal tissue sequence variants from the mitogenomes in tumour tissue, which generated the same result as using minor-allele-frequency-based phasing.

Nuclear genomic DNA variant calling

Nuclear genomic DNA was analysed for SNVs in short read DNA data and SVs in short- and long-read DNA data using custom R scripts. SNVs were determined as described above for mitochondrial DNA with the exception that we choose to haplotype all samples a ploidy of N = 2. Owing to the large number of small contigs in the reference genome, we called variants on the reads that mapped to the first 100 scaffolds (around 85% of the genome). We further filtered the dataset to include only samples with at least 8× coverage (Supplementary Table 4) of the reference genome and a VAF > 0.10 (Supplementary Table 6; see the analysis below for details).

We identified tumour specific variants in two ways. For each fish with a paired normal and tumour sample (n = 17 pairs >8×), we compared the allele frequency for all biallelic sites where 90% of samples had calls with at least 10× coverage and determined which alleles (reference or variant) were shared by the tumour and normal samples, found only in the tumour samples, found only in normal sample or were absent in either sample. We then counted the number of times an allele was found only in normal or only in tumour samples. For example, for a given SNV that was identified as occurring only in tumour samples, we counted the number of tumour samples that contained that SNV and determined it was shared by 1, 2, 3, 4 or n fish tumour samples. We also calculated the allele frequency of the tumour-only variants across each tumour tissue to determine the average and median VAF.

We also determined tumour-specific variants using GATK’s somatic variant caller Mutect2 (ref. ⁶¹). For each sample pair, we used the normal sample as the germline and used it to call somatic (in this case, tumour) variants in the tumour sample. We used the HL4 genome as the reference genome, allowed a minimum depth of 5 reads and allowed the caller to genotype germline sites. We then filtered the output with GATK’s FilterMutectCalls and extracted the somatic ‘PASS’ variant sites for each pair. The somatic variant sites across samples were compiled into a distinct list. We used the same strategy as above to determine whether variant alleles were present only in tumour or normal samples.

We called SVs—deletions, duplications, insertions, inversions and CNV—for short-read data using Delly2 (v.1.3.2)⁶². For long-read SV calling (deletions, duplications, insertions, inversions), long-read sequences were mapped to the HL4 draft genome using Dorado. Reads with minimum average quality of 10 and minimum length of 500 were filtered using Samtools and retained for further analyses. Owing to low coverage and a relatively high error rate, we did not call SNVs using long-read data. However, SVs were called using sniffles2 (v.2.5.2)⁶³ using the default parameters. We determined biallelic tumour-specific variants using the methods described above on SV sites that had coverage for at least 90% of samples used in each analysis.

Phylogenetic analyses

Variant trees

Phylogenetic trees were built from mitogenomic SNVs (combined SNVs and indels), biallelic nuclear SNVs that had 10× coverage of the first five contigs of the HL4 genome in at least 95% of the samples and nuclear SVs that had coverage of the first 100 contigs of the HL4 genome in at least 90% of the samples in the long- and short-read sequence datasets from deletions, duplications, insertions and inversions. For all sets of variants, phylogenetic trees were generated using R statistical software (v.4.4.2)⁶⁰ package fastreeR (v.3.23)⁶⁴ using an agglomerative Neighbour Joining method based on a hierarchically clustered distance matrix (a cosine type dissimilarity measurement) generated between the samples of the variant (VCF) file. The tree was converted to Newick format with ape (v.5.8-1)⁶⁵.

Mitochondrial trees

Phased mitochondrial genomes were aligned using MAFFT (v.7.526)⁶⁶. Maximum-likelihood phylogenetic trees were built from the alignment using IQTREE (v.2.3.5)⁶⁷ with 1,000 bootstrap replicates. SNPs were extracted from alignments using SNP-sites (v.2.5.1)⁶⁸. All phylogenetic trees were visualized with ggtree (v.3.13)⁶⁹ and treeio (v.1.30.0)⁷⁰.

Determining shared somatic mutations in human melanoma

To test how the distribution of shared tumour-specific mutations between bullhead melanomas compared to shared tumour-specific mutations in conventional melanomas, we downloaded somatic mutation calls from TCGA²¹. Notably, TCGA data are exon-only, so they capture only around 2% of whole-genome mutations, but as same exons are sequenced for each sample, this is still informative for calculating the ratio of SNVs that are unique to a single tumour, versus shared by multiple tumours. We filtered for melanomas (SKCM, n = 463) then, for each SNV that existed in this set, we counted the number of samples that an identical SNV was found in, aggregating SNVs that were found in 16 or more tumours into one bin, plotting the distribution in Extended Data Fig. 6a. To more closely mimic the parameters of our brown bullhead tumour dataset, we randomly sampled 16 melanomas; then, for each SNV that existed in this set, we counted the number of samples that SNV was found in. We ran this random sampling 100 times, plotting the distribution as box plots in Extended Data Fig. 6b. Notably, across all 100 resamplings, we did not observe a single occurrence of a single mutation shared by all 16 tumours.

Screening for microorganisms in tissue samples

We screened tissue DNA and RNA sequences for viral and other cellular organisms. Any sample read that did not map to the HL4 genome was processed used nf-core/mag (v.3.3.0)^71,72 to generate assemblies^49,73 and run classification of mobile genetic elements (virus and plasmid) using geNomad v.1.8.1 on assembled contigs⁷⁴ and classification of prokaryotic and eukaryotic organisms on sequencing reads with Kraken (v.2.1.3)⁷⁵. Kraken2 data were imported into R using phyloseq (v.3.23)⁷⁶. We examined differential abundance of taxa between tumour and normal samples for each dataset (RNA or DNA) using DESeq2 (ref. ⁷⁷). To remove false positives, we filtered out organisms with less than 50 counts across the dataset, and those that were present in fewer than a quarter of the sample subset before running the analysis.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

No restrictions apply to data availability. Genomic sequencing data are available at the NCBI under BioProject accessions PRJNA1284832 and PRJNA1284812; SRA accessions SRR34320463–SRR34320479 and SRR34328784–SRR36648244) as are the draft Genomes (PV874414.1 and GCA_051624135.1).

No restrictions apply to code availability. Bash and R scripts that were used in this study are available at GitHub (https://github.com/limey-bean/Bullhead_2025/). Most analyses were performed on the high-performance compute cluster at the University of Vermont’s Vermont Advanced Computing Center (RRID: SCR_017762). Custom assembler Pegasus is available at Zenodo⁴⁵ (https://doi.org/10.5281/zenodo.11397779) and GitHub (https://github.com/jaxlub/PEGASUS).

1. Blazer, V. S., Shaw, C. H., Smith, C. R., Emerson, P. & Jones, T. Malignant melanoma of brown bullhead (Ameiurus nebulosus) in Lake Memphremagog, Vermont/Quebec. J. Fish Dis. 43, 91–100 (2020).

   Article  PubMed  Google Scholar 

2. Metzger, M. J. & Goff, S. P. A sixth modality of infectious disease: contagious cancer from devils to clams and beyond. PLoS Pathog. 12, e1005904 (2016).

   Article  PubMed  PubMed Central  Google Scholar 

3. The Cancer Genome Atlas Research Network. The Cancer Genome Atlas Pan-Cancer analysis project. Nat. Genet. 45, 1113–1120 (2013).

4. Murgia, C., Pritchard, J. K., Kim, S. Y., Fassati, A. & Weiss, R. A. Clonal origin and evolution of a transmissible cancer. Cell 126, 477–487 (2006).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

5. Pearse, A.-M. & Swift, K. Transmission of devil facial-tumour disease. Nature 439, 549 (2006).

   Article  ADS  CAS  PubMed  Google Scholar 

6. Metzger, M. J., Reinisch, C., Sherry, J. & Goff, S. P. Horizontal transmission of clonal cancer cells causes leukemia in soft-shell clams. Cell 161, 255–263 (2015).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

7. Metzger, M. J. et al. Widespread transmission of independent cancer lineages within multiple bivalve species. Nature 534, 705–709 (2016).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

8. Yonemitsu, M. A. et al. A single clonal lineage of transmissible cancer identified in two marine mussel species in South America and Europe. eLife 8, e47788 (2019).

9. Skazina, M. et al. First description of a widespread Mytilus trossulus-derived bivalve transmissible cancer lineage in M. trossulus itself. Sci Rep. 11, 5809 (2021).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

10. Hammel, M. et al. Prevalence and polymorphism of a mussel transmissible cancer in Europe. Mol. Ecol. 31, 736–751 (2022).

    Article  CAS  PubMed  Google Scholar 

11. Michnowska, A., Hart, S. F. M., Smolarz, K., Hallmann, A. & Metzger, M. J. Horizontal transmission of disseminated neoplasia in the widespread clam Macoma balthica from the Southern Baltic Sea. Mol. Ecol. 31, 3128–3136 (2022).

    Article  PubMed  PubMed Central  Google Scholar 

12. Garcia-Souto, D. et al. Mitochondrial genome sequencing of marine leukaemias reveals cancer contagion between clam species in the Seas of Southern Europe. eLife 11, e66946 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

13. Yonemitsu, M. A. et al. Multiple lineages of transmissible neoplasia in the basket cockle (C. nuttallii) with repeated horizontal transfer of mitochondrial DNA. Mol. Ecol. 34, e17682 (2025).

    Article  CAS  PubMed  Google Scholar 

14. Sakaris, Jesien, P. C., Roman, V. & & Pinkney, A. E. Brown bullhead as an indicator species: seasonal movement patterns and home ranges within the Anacostia River, Washington, D.C. Trans. Am. Fish. Soc. 134, 1262–1270 (2005).

    Article  Google Scholar 

15. Agency of Natural Resources, Department of Environmental Conservation. Lake Memphremagog. https://dec.vermont.gov/watershed/restoring/memphremagog (2026).

16. Ní Leathlobhair, M. & Lenski, R. E. Population genetics of clonally transmissible cancers. Nat. Ecol. Evol. 6, 1077–1089 (2022).

    Article  PubMed  Google Scholar 

17. Baez-Ortega, A. et al. Somatic evolution and global expansion of an ancient transmissible cancer lineage. Science 365, eaau9923 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

18. Pye, R. J. et al. A second transmissible cancer in Tasmanian devils. Proc. Natl Acad. Sci. USA 113, 374–379 (2016).

    Article  ADS  CAS  PubMed  Google Scholar 

19. Stammnitz, M. R. et al. The evolution of two transmissible cancers in Tasmanian devils. Science 380, 283–293 (2023).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

20. Osburn, R. C. Black tumor of the catfish. Bull. Bur. Fish. 41, 9–13 (1925).

    Google Scholar 

21. Riquet, F., Simon, A. & Bierne, N. Weird genotypes? Don’t discard them, transmissible cancer could be an explanation. Evol. Appl. 10, 140–145 (2017).

    Article  CAS  PubMed  Google Scholar 

22. Martincorena, I. et al. Universal patterns of selection in cancer and somatic tissues. Cell 171, 1029–1041 (2017).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

23. Alexandrov, L. B. et al. Signatures of mutational processes in human cancer. Nature 500, 415–421 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

24. Ellrott, K. et al. Scalable open science approach for mutation calling of tumor exomes using multiple genomic pipelines. Cell Syst. 6, 271–281 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

25. Burioli, E. A. V. et al. Traits of a mussel transmissible cancer are reminiscent of a parasitic life style. Sci Rep. 11, 24110 (2021).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

26. Giersch, R. M. et al. Survival and detection of bivalve transmissible neoplasia from the soft-shell clam Mya arenaria (MarBTN) in seawater. Pathogens 11, 283 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

27. Hart, S. F. M., Garrett, F. E. S., Kerr, J. S. & Metzger, M. J. Gene expression in soft-shell clam (Mya arenaria) transmissible cancer reveals survival mechanisms during host infection and seawater transfer. PLoS Genet. 21, e1011629 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

28. Blumer, L. S. Reproductive natural history of the brown bullhead Ictalurus nebulosus in Michigan. Am. Midl. Nat. 114, 318–330 (1985).

    Article  Google Scholar 

29. Harris, J. & Bird, D. J. Modulation of the fish immune system by hormones. Vet. Immunol. Immunopathol. 77, 163–176 (2000).

    Article  CAS  PubMed  Google Scholar 

30. McCallum, H. et al. Distribution and impacts of Tasmanian devil facial tumor disease. EcoHealth 4, 318–325 (2007).

    Article  Google Scholar 

31. Frampton, D. et al. Molecular signatures of regression of the canine transmissible venereal tumor. Cancer Cell 33, 620–633 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

32. Farley, C. A., Plutschak, D. L. & Scott, R. F. Epizootiology and distribution of transmissible sarcoma in Maryland softshell clams, Mya arenaria, 1984-1988. Environ. Health Perspect. 90, 35–41 (1991).

    CAS  PubMed  PubMed Central  Google Scholar 

33. Muttray, A. et al. Haemocytic leukemia in Prince Edward Island (PEI) soft shell clam (Mya arenaria): spatial distribution in agriculturally impacted estuaries. Sci. Total Environ. 424, 130–142 (2012).

    Article  ADS  CAS  PubMed  Google Scholar 

34. Hart, S. F. M. et al. Centuries of genome instability and evolution in soft-shell clam, Mya arenaria, bivalve transmissible neoplasia. Nat. Cancer https://doi.org/10.1038/s43018-023-00643-7 (2023).

35. Weinandt, S.A. et al. Atlantic to Pacific: outbreak of bivalve transmissible neoplasia detected in hybridizing soft-shell clams and eDNA in Puget Sound. Proc. Natl Acad. Sci. USA 123, e2611852123 (2026).

36. Siciliano, M. J., Perlmutter, A. & Clark, E. Effect of sex on the development of melanoma in hybrid fish of the genus Xiphophorus. Cancer Res. 31, 725–729 (1971).

    CAS  PubMed  Google Scholar 

37. Sweet, M. et al. Evidence of melanoma in wild marine fish populations. PLoS ONE 7, e41989 (2012).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

38. Schall, M. K., Smith, G. D., Blazer, V. S., Walsh, H. L. & Wagner, T. Factors influencing the prevalence of hyperpigmented melanistic lesions in smallmouth bass Micropterus dolomieu in the Susquehanna River Basin, Pennsylvania. J. Fish. Dis. 48, e14033 (2025).

    Article  PubMed  Google Scholar 

39. Blazer, V. S. et al. Melanoma and other melanistic lesions in brown bullhead Ameiurus nebulosus from waterbodies in the northeastern United States and Canada: identification of risk factors. J. Fish Dis. https://doi.org/10.1111/jfd.70207 (2026).

40. Jin, J.-J. et al. GetOrganelle: a fast and versatile toolkit for accurate de novo assembly of organelle genomes. Genome Biol. 21, 241 (2020).

    Article  PubMed  PubMed Central  Google Scholar 

41. Chen, S., Zhou, Y., Chen, Y. & Gu, J. fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics 34, i884–i890 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

42. Lopez, M. L. D. et al. mtGrasp: streamlined reference-grade mitochondrial genome assembly and standardization to enhance metazoan mitogenome resources. Methods Ecol. Evol. 16, 668–677 (2025).

    Article  Google Scholar 

43. Li, H. Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics 34, 3094–3100 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

44. Danecek, P. et al. Twelve years of SAMtools and BCFtools. GigaScience 10, giab008 (2021).

    Article  PubMed  PubMed Central  Google Scholar 

45. Lubkowitz, J., Curd, E., Dragon, J. & Harcourt, E. PEGASUS: a comprehensive hybrid genome assembly pipeline. Zenodo https://doi.org/10.5281/zenodo.11397779 (2024).

46. De Coster, W. & Rademakers, R. NanoPack2: population-scale evaluation of long-read sequencing data. Bioinformatics 39, btad311 (2023).

    Article  PubMed  PubMed Central  Google Scholar 

47. Andrews, S. FastQC: a quality control tool for high throughput sequence data. Babraham Bioinformatics https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ (2010).

48. Kim, D., Song, L., Breitwieser, F. P. & Salzberg, S. L. Centrifuge: rapid and sensitive classification of metagenomic sequences. Genome Res. 26, 1721–1729 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

49. Lin, Y. et al. Assembly of long error-prone reads using de Bruijn graphs. Proc. Natl Acad. Sci. USA 113, E8396–E8405 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

50. Aury, J.-M. & Istace, B. Hapo-G, haplotype-aware polishing of genome assemblies with accurate reads. NAR Genom. Bioinform. 3, lqab034 (2021).

    Article  PubMed  PubMed Central  Google Scholar 

51. Vaser, R., Sović, I., Nagarajan, N. & Šikić, M. Fast and accurate de novo genome assembly from long uncorrected reads. Genome Res. 27, 737–746 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

52. Coombe, L. et al. LongStitch: high-quality genome assembly correction and scaffolding using long reads. BMC Bioinform. 22, 534 (2021).

    Article  CAS  Google Scholar 

53. Manni, M., Berkeley, M. R., Seppey, M. & Zdobnov, E. M. BUSCO: assessing genomic data quality and beyond. Curr. Protoc. 1, e323 (2021).

    Article  PubMed  Google Scholar 

54. Gurevich, A., Saveliev, V., Vyahhi, N. & Tesler, G. QUAST: quality assessment tool for genome assemblies. Bioinformatics 29, 1072–1075 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

55. Pedersen, B. S. & Quinlan, A. R. Mosdepth: quick coverage calculation for genomes and exomes. Bioinformatics 34, 867–868 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

56. Cantalapiedra, C. P., Hernández-Plaza, A., Letunic, I., Bork, P. & Huerta-Cepas, J. eggNOG-mapper v2: functional annotation, orthology assignments, and domain prediction at the metagenomic scale. Mol. Biol. Evol. 38, 5825–5829 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

57. Seemann, T. Barrnap. GitHub https://gitbhub.com/tseemann/barrnap (2025).

58. Auwera, G. van der. Genomics in the Cloud: Using Docker, GATK, and WDL in Terra (O’Reilly Media, 2020).

59. Robinson, J. T. et al. Integrative Genomics Viewer. Nat. Biotechnol. 29, 24–26 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

60. R Core Team. R: a language and environment for statistical computing. R https://www.R-project.org/ (2025).

61. Benjamin, D. et al. Calling somatic SNVs and indels with Mutect2. Preprint at bioRxiv https://doi.org/10.1101/861054 (2019).

62. Rausch, T. et al. DELLY: structural variant discovery by integrated paired-end and split-read analysis. Bioinformatics 28, i333–i339 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

63. Smolka, M. et al. Detection of mosaic and population-level structural variants with Sniffles2. Nat. Biotechnol. 42, 1571–1580 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

64. Gkanogiannis, A. & Bruls, T. A scalable assembly-free variable selection algorithm for biomarker discovery from metagenomes. BMC Bioinformatics 17, 311 (2016).

65. Paradis, E. & Schliep, K. ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35, 526–528 (2019).

    Article  CAS  PubMed  Google Scholar 

66. Katoh, K., Misawa, K., Kuma, K. & Miyata, T. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 30, 3059–3066 (2002).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

67. Minh, B. Q. et al. IQ-TREE 2: new models and efficient methods for phylogenetic inference in the genomic era. Mol. Biol. Evol. 37, 1530–1534 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

68. Page, A. J. et al. SNP-sites: rapid efficient extraction of SNPs from multi-FASTA alignments. Microb. Genom. 2, e000056 (2016).

    PubMed  PubMed Central  Google Scholar 

69. Yu, G. Data Integration, Manipulation and Visualization of Phylogenetic Trees (Chapman and Hall/CRC, 2022); https://doi.org/10.1201/9781003279242.

70. Wang, L.-G. et al. Treeio: an R package for phylogenetic tree input and output with richly annotated and associated data. Mol. Biol. Evol. 37, 599–603 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

71. Ewels, P. A. et al. The nf-core framework for community-curated bioinformatics pipelines. Nat. Biotechnol. 38, 276–278 (2020).

    Article  CAS  PubMed  Google Scholar 

72. Krakau, S., Straub, D., Gourlé, H., Gabernet, G. & Nahnsen, S. nf-core/mag: a best-practice pipeline for metagenome hybrid assembly and binning. NAR Genom. Bioinform. 4, lqac007 (2022).

    Article  PubMed  PubMed Central  Google Scholar 

73. Nurk, S., Meleshko, D., Korobeynikov, A. & Pevzner, P. A. metaSPAdes: a new versatile metagenomic assembler. Genome Res. 27, 824–834 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

74. Camargo, A. P. et al. Identification of mobile genetic elements with geNomad. Nat. Biotechnol. 42, 1303–1312 (2024).

    Article  ADS  CAS  PubMed  Google Scholar 

75. Wood, D. E., Lu, J. & Langmead, B. Improved metagenomic analysis with Kraken 2. Genome Biol. 20, 257 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

76. McMurdie, P. J. & Holmes, S. Phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE 8, e61217 (2013).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

77. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Author notes

1. These authors contributed equally: Emily E. Curd, Samuel F. M. Hart

2. These authors jointly supervised this work: Mark J. Henderson, Peter Emerson, Julie A. Dragon

Authors and Affiliations

1. Vermont Biomedical Research Network, University of Vermont, Burlington, VT, USA

   Emily E. Curd, James Lubkowitz, Lucas Milazzo & Julie A. Dragon

2. Department of Genome Sciences, University of Washington, Seattle, WA, USA

   Samuel F. M. Hart

3. Department of Biomedical Informatics, University of Utah, Salt Lake City, UT, USA

   James Lubkowitz

4. Vermont Integrative Genomics Resource, University of Vermont, Burlington, VT, USA

   Kirsten M. Tracy

5. Vermont Fish and Wildlife Department, Montpelier, VT, USA

   Matthew Bodnar, Tom Jones & Peter Emerson

6. US Geological Survey, Vermont Cooperative Fish and Wildlife Research Unit, Rubenstein School of Environment and Natural Resources, University of Vermont, Burlington, VT, USA

   Mark J. Henderson

7. Department of Microbiology and Molecular Genetics, University of Vermont, Burlington, VT, USA

   Julie A. Dragon

Contributions

E.E.C. co-led genomics study design, processed and analysed the brown bullhead genomics datal; co-led manuscript conceptualization and created all brown bullhead figures and wrote portions of the manuscript and the Supplementary Information; edited the manuscript; created and annotated the reference mitochondrial and nuclear genomes; supervised the creation of and contributed to the PEGASUS workflow and nuclear genome assembly; and contributed to genomic sample preparation. S.F.M.H. contributed to data analysis methods conceptualization to test the transmissible cancer hypothesis; analysed human melanoma data; co-led manuscript conceptualization; wrote portions of the manuscript and the Supplementary Information; and edited the manuscript. J.L. built the PEGASUS workflow and contributed the HL4 genome build. K.M.T. contributed to genomic sample preparation and data generation at UVM and provided sections of the Methods. L.M. contributed to the viral and microorganism screening analysis. P.E. carried out study design, offered fisheries biological expertise, performed sample collection and procured funding. M.B. contributed to the planning and field sampling for the genomics experiment, collected and processed tissues from 2022 reference and 2023 Lake Memphremagog fish and edited the manuscript. T.J. contributed to the planning and sampling of Lake Memphremagog. M.J.H. procured funding, contributed to study design and edited the manuscript. J.A.D. co-led the genomics study design and oversaw sequencing and data analysis, co-led manuscript conceptualization, provided portions of the manuscript and edits to other contributions, contributed to sample preparation and procured funding.

Corresponding author

Correspondence to Julie A. Dragon.

Competing interests

The authors declare no competing interests.

Peer review information

Nature thanks Beata Ujvari, who co-reviewed with Rodrigo Hamede; and Andrew Storfer and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions